I have long been convinced of the soundness of the luck argument against standard accounts of libertarianism (the qualification ‘standard’ is necessary; I believe that libertarianisms that are no more subject to luck than the best compatibilisms are possible, but I don’t want to get into that argument right now). The luck argument, as I understand it, targets moral responsibility: if an agent Xs, where X-ing is bad, and they are lucky to X, inasmuch as had luck played out differently, they would have performed some action incompatible with X (or not action at all), then they are not responsible for X-ing.
The question I am currently grappling with is why are standard libertarianisms subject to an unacceptable degree or kind of luck: that is, what condition or conditions on moral responsibility are not satisfied by an agent who is subject to responsibility-undermining luck? The way in which I set out the luck argument above is influenced by (or perhaps influences) the condition on moral responsibility I am tempted by. Call it the contrastive principle. It is intended as a necessary condition:
An agent is morally responsible for X-ing only if the event of his X-ing is the (non-deviant) upshot of his decision to X rather than Y, where X-ing and Y-ing have conflicting moral valences. An agent can satisfy the contrastive principle either directly – by deciding to X rather than Y – or indirectly; by being strongly disposed on the current occasion to X because of past occasions on which he directly satisfied the contrastive principle.
The above is probably a little obscure, so let me say just a little bit more. Moral valences are polarities; the moral valence of an act is its goodness or badness. Thus, an agent satisfies the contrastive principle directly by choosing a bad action rather than a good one, or vice-versa (actually, I think moral responsibility unlike moral goodness tracks subjective judgments; so really it is by choosing an action that he takes to be bad rather than good that an agent satisfies the principle). I want the contrastive principle to be satisfiable by agents in Frankfurt-style cases (despite my doubts about such cases). The intuitive idea is that we blame agents for choosing the bad rather than the good, and vice-versa for praise.
I have a feeling that the contrastive principle has problems. At the moment, its only a vague feeling. Want to help turn it into a conviction?